EARS: A Framework For Designing Embodied Conversational Interface Agents
To summarize and synthesize the four design phases explored in developing a framework for designing embodied conversational interface agents, we can use the acronym EARS, standing for Embodiment, Anthropomorphism, Realism, and Situation.
I. Embodiment
Three categories of embodiment are considered in embodied conversational interface design: humans, nonhuman objects and animals, and robots. When considering a human embodiment, designers should take into account the gender, race, and class signifiers inherent in the agent’s presentation. When considering nonhuman objects and animal embodiments, designers should take into account the fact that no object is neutral in its cultural context. When considering robotic embodiments, designers can reference the discipline of social robot design and human-robot interaction to create robots that have both mechanical and human features. Designers should also remember that users can not be trusted: allowing users to customize their interface agent will not necessarily produce the best agent for that user.
II. Anthropomorphism
The key features to keep in mind when applying anthropomorphism to an interface are the attribution of both physical and mental human characteristics to the interface agent. Adding specific facial features, at minimum, the addition of detailed eyes, has a profound psychological effect on interface and product design. Using existing frameworks like Kindenschema and emotional animation to make interface characters more likeable will also make users perceive them as more trustworthy, intelligent, and easier to use. When considering how anthropomorphic an embodied conversational agent should be, designers should remember that making an agent appear more humanlike and more capable than it actually is may result in frustration and disappoint with the agent’s actual abilities.
III. Realism
Making an embodied agent more realistic does not necessarily make the agent, or the design, better. Designers can subvert the Uncanny Valley effect by choosing the correct aesthetic. One way to aesthetically appeal to users is to make the interface more attractive; another is to not aim for realism in the first place, the argument for lo-fi design in high tech. As the capacity for realism in 3-Dimensional interface agents increases, designers will still be faced with the choice whether to make them as realistic as possible, and low fidelity designs such as illustrated characters may still be more effective and have less risk of a negative affect.
IV. Situation
Designers must take into account the virtual, physical, and social situation of the embodiment of a conversational interface. Designing for the agent’s virtual situation means creating a digital context, or space for the character to live in within the interface when in use and not in use. Designing for the agent’s physical situation, whether it appears on a desktop, mobile device screen, integrates with a smart speaker, or confronts the user life-sized through virtual or augmented reality, will also change the paramaters of the embodied agent’s design. Adding accoutrements of a social role to the embodied agent’s design will also make users more comfortable by situating the agent within an explicit social context.
Using the EARS framework is one method for designers to tackle the many questions that designing embodied conversational interface agents open up as to agents’ usability and the future of a more human design for technology. Considering the technological trend towards natural language interfaces through developments in artificial intelligence and messaging apps, and the prevalance of personal assistant devices and chatbots being applied to hundreds of different industries, the natural tendency for us to interact with technology as we interact with other humans is bound to increase. Designers will have a key role in the development of the human face of technology, which shapes how we relate not only to devices, but also to each other.